Welcome to KeepUrAi: Company AI That Stays on Your Server

The artificial intelligence revolution is here, but for many small and mid-size companies (KMU), there is a growing tension between wanting to use AI and needing to protect sensitive data. KeepUrAi is built on the premise that you should not have to choose one over the other: your models, your data, and your infrastructure can stay under your own control.

The Problem with Cloud-Only AI

When organizations adopt public AI services without thinking it through, they often overlook one thing: data leaves their control. Every prompt, every document, every query sent to a cloud AI service travels to servers owned and operated by a third party. That creates a few recurring problems.

Data Sovereignty

For organizations handling sensitive data — healthcare, finance, legal, or anyone bound by professional secrecy — sending that data to an external AI service raises real questions about who can access it and where it ends up. GDPR and Switzerland's revDSG both require a clear answer to that question; a public AI service outside your control makes it harder to give one.

Intellectual Property

When proprietary information flows through a third-party service, it becomes part of someone else's infrastructure. Trade secrets, internal documents, and competitive information carry a real exposure that many organizations would rather avoid.

Unpredictable Costs

Cloud AI services typically charge per token or per API call. As usage grows, costs can scale in ways that are hard to plan for.

Dependency

Relying on an external service means depending on that provider's uptime, network performance, and capacity — a dependency that matters more the more central AI becomes to daily work.

The Case for Running AI Yourself

A real shift is underway: modern open-source language models, served through tools like Ollama, now run effectively on infrastructure a company controls. This is not a compromise on capability — for most business use cases, it is a genuinely practical option, and it brings some concrete advantages:

Data control on-premise — in the on-premise editions, your data stays on your own servers by design.
Predictable costs — infrastructure you run yourself has a cost structure you can plan around, instead of usage-based billing surprises.
Low latency — local inference means the round trip stays inside your own network.
Air-gapped deployment — for the most sensitive environments, full network isolation is possible with the on-premise editions.

What KeepUrAi Brings to the Table

KeepUrAi was built to make this practical for organizations that do not want to run their own AI infrastructure team. It provides:

Straightforward Deployment

Deploy on your infrastructure as Debian/RPM packages, Docker Compose, or a self-managed Kubernetes setup — whichever matches how you already run your systems.

Security Built In

Role-based access control, audit logging, and encryption at rest and in transit are part of the base deployment, not an add-on. KeepUrAi holds no third-party security certification today; where it aligns with a framework like GDPR or revDSG, that is stated plainly rather than implied.

Model Flexibility

Choose from a curated set of open-source language models served through Ollama, or bring your own. Fine-tune on your own data without sending that data to an external party.

Seamless Integration

REST APIs and an MCP (Model Context Protocol) interface make it possible to connect KeepUrAi to your existing applications, workflows, and AI-agent tooling.

Who Benefits from Keeping AI In-House

Fiduciaries and Independent Asset Managers

Client confidentiality is the business. KeepUrAi lets these firms use AI for document review and research while keeping client data inside their own infrastructure.

Law Firms

Attorney-client privilege demands confidentiality. KeepUrAi allows firms to use AI for document review, contract analysis, and legal research while keeping privileged information under their own control.

Healthcare Practices

Patient data deserves particular care. With the on-premise edition, healthcare organizations can use AI for documentation and research support without that data leaving their own servers.

Research and Engineering Teams

Teams working with unpublished findings or proprietary designs can use AI for analysis and drafting while keeping that material off third-party infrastructure.

The Path Forward

The question is not whether SMEs will adopt AI — most already have, in some form. The real question is whether they do so in a way that protects their data and serves their long-term interests, rather than accumulating a dozen disconnected tools.

Get Started

Explore the documentation — architecture, deployment options, and supported models are described at /doc.
Try the hosted edition — qcg-ai.ch is a Swiss-operated instance of KeepUrAi you can log into directly, no deployment required. GPU inference for chat and image generation on the hosted edition runs through external providers (Ollama Cloud, Comfy Cloud) — the "stays on your server" guarantee above applies to the on-premise editions.
Talk to us — contact us to discuss your own deployment.

Welcome to KeepUrAi — on PrivateStudio, made by INSS in Zürich.

Have questions about self-hosted AI infrastructure? Contact our team or read our security best practices article.